module change_detection.tornado.adwin

Adaptive Windowing Drift Detection Method.

The source code was adopted from tornado, please cite:

The Tornado Framework By Ali Pesaranghader University of Ottawa, Ontario, Canada E-mail: apesaran -at- uottawa -dot- ca / alipsgh -at- gmail -dot- com

Original Paper: Bifet, Albert, and Ricard Gavalda. "Learning from time-changing data with adaptive windowing." Published in: Proceedings of the 2007 SIAM International Conference on Data Mining. Society for Industrial and Applied Mathematics, 2007. URL: http://www.cs.upc.edu/~GAVALDA/papers/adwin06.pdf

Copyright (C) 2022 Johannes Haug.

class Adwin

Adwin Change Detector.

method Adwin.__init__

__init__(delta: float = 0.002, reset_after_drift: bool = False)

Inits the change detector.

Args:

  • delta: Todo (left unspecified by the Tornado library).
  • reset_after_drift: A boolean indicating if the change detector will be reset after a drift was detected.

method Adwin.detect_change

detect_change() → bool

Detects global concept drift.

Returns:

  • bool: True, if a concept drift was detected, False otherwise.

method Adwin.detect_partial_change

detect_partial_change() → Tuple[bool, list]

Detects partial concept drift.

Notes:

Adwin does not detect partial change.

method Adwin.detect_warning_zone

detect_warning_zone() → bool

Detects a warning zone.

Notes:

Adwin does not raise warnings.

method Adwin.partial_fit

partial_fit(pr_scores: List[bool])

Updates the change detector.

Args:

  • pr_scores: A boolean vector indicating correct predictions. 'True' values indicate that the prediction by the online learner was correct, otherwise the vector contains 'False'.

method Adwin.reset

reset()

Resets the change detector.

This file was automatically generated via lazydocs.